Synthetic fast food? Auditing generative AI's representations of European states
ENThis article explores how Generative AI visually and textually represents national identities and cultures, focusing on depictions of selected Western and Eastern/Central European states. Patterns of bias embedded in AI-generated depictions of everyday life, work, family, national character etc. are identified. The article reveals a fragmented vision of “synthetic Europe”: Eastern/Central European countries are predominantly portrayed through rural, traditional, and often backward-looking imagery while Western Europe is depicted as urban, albeit in a romanticised and historicised ways. These patterns reflect and reinforce existing stereotypes, historical hierarchies, and power disparities. The findings highlight the potential role of the “coded gaze” in perceptions of the continent and likely trajectories for producing and perpetuating bias through an increasingly shared human-AI agency. The article serves as an illustration case of the potential of AI auditing that moves beyond technical compliance to address the normative and cultural dimensions of bias. Keywords: AI bias, auditing, algorithmic gaze, Generative AI, stereotypes.
- Celebrating March 8: a failed attempt at deSovietization?. Journal of Baltic studies 2021, 52, 1, 43-59.
- Jaunimo požiūriai į savo pačių, tėvų ir senelių šventes: mėgstamiausia jaunimo šventė. Lituanistica 2016, 1, 45-60.
- Miesto samprata Lietuvos etnologų ir antropologų dabuose. Res humanitariae 2018, 24, 270-288.
- Moters diena. Šventės šiuolaikinėje vilniečių šeimoje Vilnius: Lietuvos istorijos institutas, 2016 P. 189-221.
